Chest
CommentaryStatistical QuandariesDo CIs Give You Confidence?
Section snippets
The Ubiquitous p Value
The use of P values has a long history, dating back to seminal work by Fisher and Pearson in the 1920s. In its simplest form comparing two groups, the P value is “the probability of observing a result at least as extreme as the observed result if in truth there is no difference between the groups.”2 A small P value (P < .0001) says that it is unlikely that a difference this large could arise by chance (< one chance in a thousand), so it is likely a real effect.
It is critical to keep in mind
The Less Ubiquitous CI
CIs turn this whole thing on its head. We begin with the observed difference, call it “d.” We superimpose a normal distribution centered on d, then take the same 1.96 SEM and add and subtract it from d. The upper and lower limits define the 95% CI, where we can be confident that the true difference lies within these boundaries 95% of the time. So a narrow CI indicates that the difference has been estimated with a high degree of precision, and a wide CI says that there is much more uncertainty
Putting It Into Practice
It should be evident by now that despite the rhetoric, it really is not the case that one is holy and the other is evil. They really are alternate routes to expressing very similar ideas, and as a consequence, you cannot really say that the CI should be used under some circumstances and the P value under others. Indeed, one article we came across in CHEST (and we are quite sure it is not unique) had the best of both worlds:
Independent factors associated with an accelerated decline of lung
Acknowledgments
Financial/nonfinancial disclosures: The authors have reported to CHEST that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.
References (4)
- et al.
Correction for multiple testing: is there a resolution?
Chest
(2011) - et al.
Factors associated with lung function decline in adult patients with stable non-cystic fibrosis bronchiectasis
Chest
(2007)
Cited by (0)
Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (http://www.chestpubs.org/site/misc/reprints.xhtml).